A Mutated Salp Swarm Algorithm for Optimization of Support Vector Machine Parameters

Abstract

Support Vector Machine (SVM) is typically a supervised learning algorithm that carefully examines input and identifies distinct patterns. The function of SVM classifier relies on adjusting or controlling of kernel and penalty parameter values. Nature Inspired Algorithm helps to solve the natural problems and has been attracting considerable attention due to their better performance. Salp Swarm Algorithm (SSA) is a Nature Inspired Algorithm (NIA) which is used to control the finest SVM parameters value. To improve exploration capability of SSA, mutation method is developed to find the optimal value for kernel parameter and penalty parameter. The preliminary result indicates Mutated SSA with SVM increases classification accuracy than simple SSA with SVM.